Signature Prediction and Uncertainty Analysis for Radar-based MDA Applications

Incorporating the possible presence of delayed radar returns due to traveling waves, cavity resonance, and other physics-based features in the automatic object recognition (AOR) database can greatly improve the odds for successful discrimination. Mapping out the full range of variation in the radar return, over angle, frequency, and object variability, would seem to require vast computational resources, if the return is to be predicted with useful accuracy. Radar cross section (RCS) can vary sensitively with small changes in the various parameters, applying standard interpolation techniques to obtain the required resolution becomes impractical. Under the SBIR Phase I contract HQ0006-09-C-7150, HyPerComp has made significant strides in advancing the state of the art in physics-based computational electromagnetics (CEM) in the following areas: 1) high-order accurate discontinuous Galerkin (DG)-based algorithms, 2) uncertainty estimation based on a chaos polynomial expansion procedure, 3) a reduced order-basis method (RBM) for greatly minimizing the computational burden of generating large data domes, and 4) exploiting the advances in graphic processing unit (GPU) computing for dramatic speed-ups in computation over conventional central processing unit (CPU) type computing. In the Phase II effort, the end goal is to advance the state of the art in these areas to provide user-friendly, turn key computing platforms for routine computation of physics-based, accurate, full wave solutions for MDA objects of interest at greatly reduced computational run times (four to six orders of magnitude reduction) over current practice.